Hacker News new | past | comments | ask | show | jobs | submit login

Capital costs vs. marginal costs.

You're going to build a $100M desalination plant and run it for three hours a day? That's a ton of money sitting idle most of the day, far more than what is recovered with zero operating costs.

(This is called the utilization factor -- how long a piece of equipment is used vs. staying idle)

Ideally you want useful processes with low capital costs and expensive marginal/energy costs. Desalination is not one of those.

Desalinization isn't a good example for the reasons you say, but theptip writes below that heating water in residential and commercial settings might be better. Since water once heated stays hot for a good while in the insulated storage tank. Also buildings can be heated with hot water instead of forced air, typically through pipes in the floor. Which is nicer because it keeps your feet toasty and doesn't dry out the room so badly.

A freezer is possibly another such use case, if it keeps the temperature to within +/- X degrees of the set temperature, it can delay kicking in during periods of expensive electricity, and over-cool during periods of cheap electricity.

I could load the washer or dryer and program it to run it when it's cheapest if I'm not in a hurry.

Charging an electric car could work the same way.

Air conditioners are a huge deal, my single biggest electricity consumer. In a world where things were smarter they would chill coolant/water/ice during times of cheap electricity, and that would be used to cool during the hot evenings.

The problem is I have no way to take advantage of any of that. The electric grid isn't giving me feedback on the spot price of electricity (and in many places they don't even break it down on the bill, showing just one value for $/kwH for the entire month - that's what my bill looks like.)

Even if that problem is resolved, my appliances are too dumb to take advantage of that information to optimize my costs.

The grid needs to convey the realtime pricing information to customers, and then smarter appliances will be developed to take advantage of that information. The market may be able to do a surprisingly effective job of moderating demand itself if both the information and the means to apply it are available.

> but theptip writes below that heating water in residential and commercial settings might be better.

When I worked at Ørsted, we did this all the time. We had a huge electric kettle, which we used to produce district heating water.

But there are many more things you can do with surplus energy. You can make methane gas from pure CO2 and hydrogen when you apply electric power as a catalyst.

The methane can then be eaten by bacteria to provide protein powder to be used as supplements to feeding livestock.

Or, we can yank CO2 out of the atmosphere and store it as ethanol. I think this was discovered last year or so? Ethanol is a really good "battery", and it only discharges as much CO2 as was captured by the surplus energy used to make it, making it a clean fuel that only exhausts CO2 and sterile water.

A banal long term way of storing energy is to pump water up to some high ground. Whenever you need it, use a turbine to get the energy back.

This is exactly how the 6th largest power plant in the UK works (but the turbine is reversed to become the pump)



There's a number of them all over the world. Surprisingly simple idea and amazingly powerful storage.


The train goes up, the train goes down: a simple new way to store energy https://www.vox.com/2016/4/28/11524958/energy-storage-rail

This sounds good to me and it claims have 86% efficiency.

considering how much trains weigh and how fast they move, perhaps the electrical trains are already being used as "linear flywheels" ?

Since kinetic energy = 1/2 m * v ^ 2 , the a slight change in velocity on the trains that are already moving fast at any point in time, could store a lot of energy (i.e. for 2 identical weight trains, from 0km/s to 1km/s is a smaller change in stored kinetic energy than from 100km/s to 101km/s ! [in fact the latter increase of 1 km/s stores 201 times as much kinetic energy than the former: ((101 * 101) - (100 * 100)) / ((1 * 1) - (0 * 0))]

Now that I think of it, this could probably explain why our local trains are suffering more and more irregular arrival times :) but why would it be kept secret or hidden in plain sight? perhaps all the negative news about negative prices for renewable energy during energy flood is just manufacturing consent to keep price hikes for the plebs palatable, or a kind of white lie to offset their airplane travels...

to be honest I'm not very impressed by the trains up/down proposal:

The trains move on a track 5.5 miles long at an inclination of 8 degrees. thats a height difference of sin(8deg) * 5.5miles * 1.609344km/mile = 1.232 km, now it may be hard to find a steep cliff 1.2km high, but you could use a smaller cliff and heavier weights, think of the steep section at the start of an amusement ride (they will probably be better equipped with safety for such systems anyway since they are used to designing crazy rides for human consumption). no need for train and electrical tracks 5.5 miles long, since the motor can reside on the top part of the cliff/hill...

Interesting, I wonder how long a rail system would need to be in order to be able to pull energy from tidal effects with the moon.

this reminded me of underwater energy storage pilot:


Fascinating. When you pump the water out I wonder what it's replaced with? What holds back to pressure of the ocean at 700 meters depth? The article was very light on details, unfortunately.

My understanding is that they do not replace it with anything.

I'd assume the only thing they look at is not to go above the pressure differential that the dome/sphere can sustain.

That requires big water deposits up in the mountains, along with a dam.

Norway is having great success with this strategy on their hydro plants.

The UK pretty much exhausted its stock of good, easy, large sites with Dinorwig.

Amongst the best and ost efficient energy storage options, but very limited suitable sites.

The Balkans region, along the Adriatic coast, offers an interesting prospect: using the sea as a lower basin with mountaintop reservoirs. This is a rare topology, particularly near large populations.

What the results of localised salinisation might be is a concern though.

One more example.

Bath County Pumped Storage Station in the US. Huge place.


I was at a tour of Skærbækværket many years ago, and as I remember it, when you produce power, you need to be able to control the voltage somehow. If there is a surplus of power and you cannot turn down production, you need to dump it or the voltage will go up, destroying electrical systems. One of the simplest ways to do this is to heat water, and when you have well established district heating it's almost a no brainer what to do.

It's a nobrainer, but the electric kettles are rather expensive and are usually therefore only bought to ensure production capacity in the case that the powerplant's own kettle malfunctions.

Besides, district heating produced on power from the grid is taxed as if produced on coal, as the power has no traceability. This makes it really expensive, unless you are directly hooked up to a solar or windfarm.

That's the law in Denmark at least.

PS: I worked in the administrative building just next door to Skærbækværket! :)

> district heating produced on power from the grid is taxed as if produced on coal, as the power has no traceability.

Sounds like there is room for improvement. E.g. an agreement with an energy provider and tracking when it was used on demand as a sink for excess energy instead of base-rate heating.

Ahh, it doesn't make much sense to install it away from a powerplant. It captures waste heat and electricity, where as one elsewhere would only be able to capture excess electricity.

>You can make methane gas from pure CO2 and hydrogen when you apply electric power as a catalyst.

You can make methane gas from pure CO2 and hydrogen period. The reaction is exothermic, it's just the Sabatier method. You could however use electricity to split water into hydrogen and oxygen.

> We had a huge electric kettle, which we used to produce district heating water.

...and you thought you were heating with renewable energy. You were not. Much of the "surplus" power on the grid comes from coal plants (especially in Denmark!) that can't throttle down, so someone burns coal, converts the heat to electricity at an efficiency of 40% or so, and you turn it back into heat.

> ...and you thought you were heating with renewable energy.

No we didn't. As a mathematical modeller and software developer for their inhouse production optimization software, I was perfectly aware of what was happening.

> Much of the "surplus" power on the grid comes from coal plants (especially in Denmark!)

Most of the power plants owned by Ørsted are bio-converted and runs primarily on wood chips and pellets. Unless you are talking about surplus energy imported from Germany, what you say simply isn't true.

Besides, the huge amounts of surplus energy that often came from germany were caused by their massive open sea wind farms, making the energy pretty green.

Ørsed operated almost exclusively combined heat and power plants, meaning that they can produce heat and power concurrently. The utilization of energy was well above 90%, when we ran the plants this way, because we cooled the plant with the district heating water, instead of sea water. The theoretical maximum is ~98%.

We also never planned for pure power production, only to turn it into heat again. That would be monumentally stupid.

We had many, highly skilled engineers and power traders, and they absolutely knew what they were doing.

So it seems many coal burners in Denmark have switched top biomass in the past few years. I didn't know that. But someone burning wood to make electricity and someone else wasting the electricity in a resistance heater still doesn't sound like a solution, it sounds like insanity. (Not necessarily pointing fingers at you personally. Negative prices lead to insanity.)

A great example of this is ice bank refrigeration, which literally uses a giant block of ice to cool air in a building.


This is insightful. The solution is probably to have a spot market for electricity that is accessible via an API so that smart AC or fridge or electric car charger can make use of it.

This will encourage market for devices that can utilize the lower spot prices dynamically and keep the things more efficient overall (in steady state).

We've got these in New Zealand - once you have a "smart meter" (which I believe most homes now have), then you can pick an electricity retailer like https://www.flickelectric.co.nz that bills you according to the current price. Flick provides an API to allow you to turn on/off loads when you see fit.

Even before smart metres, NZ has had a ripple control system since the 1950s which sends signals over the electric lines to your circuit breaker so the utility could remotely turn domestic hot water heaters off during periods of high demand. The technology is pretty simple, you could reverse it to turn appliances on when power is below cost.


Yes, in my house that is implemented as a ripple control "receiver", which switches on a circuit that has its own meter. The receiver is a rather simple thing I think; basically a high pass filter connected to the mains, with the filter output driving the coil in a relay. It audibly hums when ripple is on. Every couple months, the meter reader comes by and records the total amp*hours consumed at "night rate" and "normal rate".

So, it's important to note that the old system doesn't allow one to run the heat pump both whenever it's wanted, and also on the lower price ripple-controlled power. Similarly in the other direction, it doesn't make sense to use ripple control to decide when to feed back in to the grid if you have generating capacity from PV or whatever.

Do they have individual control per outlet? Those loads namely ask quite a bit of power. The last mile with respect to control in the sense of a smart switch or plug per device is the most expensive part till now. That, assuming you can for internet access use the smart meter connection or the consumer's WiFi.

It seems people reuse the API which is used by the app. An example: https://www.npmjs.com/package/flick-electric-api

So it provides pricing info, not any home automation.

that's brilliant

Some electricity utilities offer this as a sort of service to their customers, where they detect high grid utilization days and notify customers they can save $ by reducing consumption on those days. Here's PG&E's page about their version: https://www.pge.com/en_US/residential/rate-plans/rate-plan-o...

It's not "smart" or automated, but it's a baby step.

This concept is called Demand Response and is quite prevalent. Consumers receive a discount for reducing their load at peak times.

This frequently targets very high demand customers rather than residential for the simple fact that it's easier to ask a company to delay starting a single piece of machinery than to reduce the load on an equivalent but large number consumer appliances like refirgerators.

Doing this at a residential level will require significant costs to support at a residential scale. Technology is the key in solving this problem.

I've been signed up with PECO for years in Pennsylvania for their SmartSaver program that targets my residential AC during the hottest summer months. I get a bill credit every month for participating.


Instead of API, they could encode relative price directly into the power source.

It’s not an encoded price but some smart grid connected devices do use the frequency as an indicator of the grid has lots of demand (slightly lower frequency, generally higher price) or lots of supply (higher frequency, generally lower price).

Rainforest Automation sells a little Linux appliance that reads smart meter data from your energy meter. Some providers, (including mine - ComEd) provide the spot price via that meter. The Rainforest device has an api both cloud & local.

Whenever I get a couple of bored minutes I go and update a little go library I have to work with it https://github.com/kklipsch/reagle

Easier solution: The net frequency is already used for load control. Require consumer devices to react to it by law. No internet or setup needed

There is a program to provide that feedback. I already own an enabled water heater for example.

https://smartgrid.ieee.org/ http://www.whatissmartgrid.org/smart-grid-101

This is the only valid reason I have heard for having smart household appliances.

You made me laugh, but yeah I can't think of another good reason for my refrigerator to be "smart".

Heating a big, isolated water tank was pretty common when I was working in the industry 25 years ago. Power was much cheaper during night so you used that to heat the water tank and daytime you used that water to heat the building. Lately this has been available to private customers also, price by the hour. (Sweden)

Check out https://mixergy.co.uk which recently came out with a nice solution.

Yep and yesterday FullyCharged released a vid about it being on trials - https://www.youtube.com/watch?v=z1Z4JCoPAGc

An interesting note from the video is how they said they could detect the grid frequency from the plug to detect times of over supply and (in theory) that would be the right time to absorb some of that excess.

Some other interesting benefits (mentioned in the comments) about how it can periodically cycle the heat better than traditional water tanks to kill off bacteria.

That's where I got it from. Was likewise intrigued by the frequency remark, interesting stuff.

Storage heaters are common in many parts of Europe: https://en.wikipedia.org/wiki/Storage_heater

They already work on this principle, more or less: heat up overnight when electricity is metered cheaper, then discharge during the day.

There are several companies working on making "smarter" versions that can switch on and off in response to real time data.

> The problem is I have no way to take advantage of any of that.

Some metropolitan areas have residential level demand response programs that address the obverse issue: when there is too little power, participating opt-in residential electricity customers with a qualifying Internet-connected thermostat would see their HVAC systems reduce power consumption through adjustment of the temperature set point. I suspect most of them use OpenADR [1].

If my skim of the OpenADR specification (requires free registration) is off, and we can't use it to pass pricing signals through the EIEvent/Quote/Report/Avail/Opt services, then we can use the OASIS Energy Interoperation parent specification, though that seems much more heavyweight and baroque to work with, and likely a harder sell to utility organizations to adopt. The utilities would need to expose pricing history as well, so longer-term planning by consumers can be performed through predictive trend analysis.

The communications and protocol infrastructure is there to convey the information to your residence, but the back-end at the utility side is an open question, and likeliest hardest to hook up. If I had access to near-zero cost electricity when they're trying shed load, then I would use it for drying clothes, chilling a pool (thermal mass) during hot months, heating a pool (thermal mass) during cold months, baking and pressure cooking while running a kitchen A/C at full blast, etc. All of these activities either use capital equipment already paid for, or very cheap to acquire to add to my existing equipment stock (like heat exchanger to thermal mass and even a brine tank).

[1] https://openadr.memberclicks.net/

I wonder if a less efficient desalination plant would be cheaper. The modern efficient plants use RO with pressure exchangers, I think. Maybe a flash distillation plant or an RO plant without pressure exchangers would be economical to operate intermittently.

But desalination may be a poor example in general. It’s not actually very energy intensive, at least compared to the value of potable water in an urban location.

Yeah, if I was gonna build a desalination plant built to use <$0 cost electricity, I would essentially build a giant (distillation) still. I'm not a process expert in flash desalination, but it looks similar to (maybe identical to) a still.

Just a big resistor in a big tank (or pond), a big fume hood and coil to another tank. Capital cost can probably be minimized a whole lot.

Seeing as "boiling the ocean" is the standard business phrase used to dismiss an overly ambitious and likely impossible to execute plan [1], there would be something beautifully ironic about a startup whose explicit goal is to solve a real world problem by literally boiling the ocean.

[1] http://workingwithmckinsey.blogspot.com/2013/11/Avoid-Boilin...

You can use sun light directly for that. Just paint your boiler black, maybe concentrate some sunlight with cheap mirrors.

This works well when you have a lot of sun. Instead you might have a lot of wind.

Also, you only want to run the desalination plant when the electricity price is too low ("below zero" with the transmission included), so you want generators for most of the time, when you actually sell the power.

Yea but not at night.

The entire point of this discussion is to figure out how to use electricity only during the day.

Even for RO the big expenses are running the pumps and fouling of the membrane, caused by high concentration of salt due to the high pressure used. If you weren't worried about throughput they last a very long time. You could use cheap electricity to pump water to a reservoir and use gravity as the pump for cheap RO. It isn't any different from any other source of off grid storage.

What about using the energy for various forms of carbon sequestration? Making biochar [1] for example? An area's biodegradable waste could be collected according to existing garbage schedules, and the grid's excess electricity used to create biochar from it at garbage/recycling sites.

[1] https://en.wikipedia.org/wiki/Biochar#Production

A desal plant is a useful thing and can be run around the clock off traditional energy sources. The "free energy" hours would help lower the costs.

I can even imagine a SETI-like application where people who over-generate power are able to donate it to causes of their preference...

But if you are using traditional energy sources to run it all the time, it is no longer 'solving' the problem of burning off excess energy during peak renewable times.

Someone is having to build a lot of highly wasteful, redundant infrastructure.

Rational cryptocurrency mining firms can use the excess (unstorable) energy by converting it back to money (while the sun shines and the wind blows).

Money > Energy > Money

> Someone is having to build a lot of highly wasteful, redundant infrastructure.

We're nowhere near having the energy infrastructure necessary to support everyone having an electric vehicle yet.

Energy storage is key to maximizing returns from renewables and minimizing irreversible environmental damage.

Not hours per day. But if your climate has wet winters and dry summers you might choose to build a desalination plant that you only really need to run for three months per year.

And then you have the plant sitting there for most of the year, able to help the power grid. So the example makes sense.

For something like a desalination plant, I wonder if that would be a good candidate for pumping water up a hill/tower/something, and then using the stored energy the other 21 hours per day.

I can't imagine that the capital required for "water boiling equipment" would be very high. Metal tanks, pumps, heating unit, etc.

Yes, but you are in one sense "competing" with other solutions (such as batteries) that face the same utilization factor problem.

Yes, depending on the timescale considered, the utilization factor of the plant should be weighed against the utilization factor of flood energy. In the long run it should pay itself back.

> In the long run it should pay itself back.

So in financial terms wouldn't this mean a rather weak profit margin? Plus the relatively high-risk that flood energy might not always result in free or negative energy prices gives it a bleak risk-to-profit ratio.

In other words, there would always be something more utilizable to be done with excess energy (e.g. mining bitcoins etc. and buy water with it when you need it).

PS: But I like the main point you make, to take in more aspects when calculating efficient renewable energy systems. Not a crypto-investor (or even believer).

Right, I explicitly mentioned not really believing desalination to be a good example, I was using it more as a fake illustrative example.

In hindsight I should have ended with posing the following question: how do we make a simple guide for factory/etc operators/consultants to recognize such steps in their production line?

Desalination is a bad example, because you need a complete new plant, while it is conceivable that other products have one or more energy dense steps.

What should such instructions look like?

Follow the product through your production line, and at each step measure the energy consumed per part at the step, and the volume per part when efficiently stacked. Also note if the step is automated or needs a human, if it is human, check if it can be automated.

If it is automatable, and the energy density for the step is high enough, calculate the cost for changing the production line, allocating sufficient storage, and possibly automating a certain task, and parallellizing the step such that it can be run during energy flood. Then calculate through the price difference how fast it pays itself back.

Mention the importance of actually measuring the energy consumption, instead of just reading off a machine specification of wattage, and cycle time.



The banks that wish to invest in cheapening such a step could train consultants and send them to factories interested in good deals. I.e. any potential profit is shared between company and bank, and the bank takes the risk (hence has the motivation to do the calculation properly, and will have best knowledge through experience of suitable step energy densities)

I think the core of your idea would work well in theory; a much simpler example would be heating water, which both residential and commercial buildings do quite a lot. Once you have hot water, you can keep hold of it for quite a long time in a properly insulated tank.

Or at a higher level of abstraction, instead of us trying to figure out which "energy dense step" is the right one to target, instead the grid can provide a real-time price signal that flexible loads can use to get a cheaper overall rate by selectively turning on when the demand is lower.

This stuff is pretty well discussed under the umbrella of "smart grid" technologies: https://en.wikipedia.org/wiki/Smart_grid#Market-enabling

In the oil market, stabilization of price occurs only because a handful of countries (really just Saudi Arabia) maintain extra capacity. They don't run their capital assets at 100%. This is where capitalism fails. You need one absolute authority, some autocrat who doesn't care that their processing plant isn't running at 100%, who is able to react instantly without making new capital improvements. This is the place for a state-owned market participant, either on the supply or consumption side.

This idea may be outdated. USA has been considerably undermining Saudi's traditional role of swing producer. Not through their centralised market control, but by having many intensely capitalistic shale operations, each with profitable capacity at a different price-point. Each one quickly toggles on/off when the market price crosses its own threshold.

just because there is fluctuation, doesn't mean there's nothing stabilizing it.

OPEC without a doubt decides output and price (with noise). They can hold the economy hostage nearly unilaterally and have for decades - see the 70s supply shock.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact